Premise: One of the best-documented ecological responses to climate warming involves temporal shifts of phenological events. However, we lack an understanding of how phenological responses to climate change vary among populations of the same species. Such variability has the potential to affect flowering synchrony among populations and hence the potential for gene flow.

Methods: To test if an earlier start of the growing season affects the potential for gene flow among populations, we quantified the distributions of flowering times of two spring-flowering plants (Trillium erectum and Erythronium americanum) over six years along an elevation gradient. We developed a novel model-based metric of potential gene flow between pairs of populations to quantify the potential for pollen-mediated gene flow based on flowering phenology.

Results: For both species investigated, earlier onset of spring led to greater separation of peak flowering dates across the elevation gradient. For T. erectum, but not E. americanum, this was also associated with a reduction in potential gene flow.

Conclusion: Our study suggests that climate change could decrease gene flow via phenological separation among populations along climatic gradients. We also provide a novel method for quantifying potential pollen-mediated gene flow using data on flowering phenology, based on a quantitative, more biologically interpretable model than other available metrics.

Two elevational transects, approximately 800m apart, were established in 2013 on the east-facing slope of Mont St-Joseph within the park (see Lajoie and Vellend 2015). Each transect comprises six 26 m x 4 m plots (104 m²), with 55-85 m of elevation between plots. Plots were established only in areas under full tree canopy to avoid disturbed habitats and wetland areas. Our focal species were selected given their high abundance and presence across most of the elevational gradient. The focal species were present in five plots per transect (absent at the highest elevation), covering an elevation gradient from 697 m to 951 m a.s.l. The geographic distance between plots varied from 123m for adjacent plots up 914m for the two most distant plots on different transects.

Each plot was divided into 26 2m × 2m subplots in which the number of flowers of Trillium erectum and Erythronium americanum was recorded approximately every 4 days throughout the flowering season for six years, 2013-2018 inclusive. Flowers were included in the count from the start of flower opening to the start of flower senescence (identified by signs of drying, wrinkling, or discoloration of petals or sepals). Since flower abundance decreased with elevation, for high-elevation plots (850 m and higher), we also recorded individuals within less than 2 m from the plots to obtain enough data to determine the flowering start, peak and end; the number of individuals was standardised to correspond to the density of individuals in an area of 104 m² (the area of one plot).